haziness of the media in a large proportion of diabetics made it difficult to visualize the capillaries in this study. Although, in Fig. 6 B, an arterial injection of fluorescein has shown up zones of capillary closure alongside arteries and scattered small areas of capillary closure (Ashton, 1953), we have been able to discern this phenomenon in only three patients after intravenous injections; usually it is not until microaneurysms and fine new vessels develop that we can show up lesions in the capillary bed with the present intravenous technique. Besides providing fine anatomical details of the lesions in diabetic retinopathy, fluorescence photography gives information on the permeability of vessels. A proportion of the microaneurysms, segments of the new vessels in the retina, the new vessels projecting into the vitreous and in the retinitis proliferans, all show leakage of fluorescein. It is not surprising that these vessels have abnormally permeable walls because of the alterations in the basement membrane of the endothelium in the microaneurysms (Bloodworth, 1963) and the defective adventitial layers of the new vessels (Ballantyne, 1946; Gartner, 1950). Hodge and Dollery (1964) have shown that on equilibrium dialysis up to 80% of fluorescein in plasma is bound to the proteins. They infer that this protein-binding is easily reversible because of rapid loss of fluorescein from the blood into the extracellular fluid after an intravenous injection. While it is certain from the photographic studies that free fluorescein (molecular weight of 361) is leaking from these microaneurysms and abnormal vessels, it is possible that fluorescein in the protein-bound form is leaking out as well. Abnormally permeable vessels could explain the retinal oedema and vitreous haze encountered in some patients with diabetic retinopathy. Retinal oedema, for example, was observed ophthalmoscopically in the young girl whose retinal photographs (Fig. 6 A and B) show numerous fine new vessels with leaking segments. Viterous haze occurs in patients with a rete mirabile or retinitis proliferans, as in the patients whose fundi are illustrated in Figs. 8 and 9. It is often unrelated to previous vitreous haemorrhage, it fluctuates throughout the day, and it is easily discerned by patient and clinician. Continued leakage of plasma proteins from the vessels of the rete mirabile -and retinitis proliferans may cause this turbidity of the vitreous. Information about the circulation rate in the retinal vessels is restricted to direct comparative observations of different vessels in the same field after each injection of fluorescein. We are able to recognize only prolonged filling of some microaneurysms and rapid shunting and delayed drainage of certain vessels, but only relative to neighbouring microaneurysms and vessels. These preliminary studies have shown that fluorescence photography can reveal the large proportion of small-vessel abnormalities hitherto invisible in the living patient, and can also demonstrate the important functional abnormality of enhanced permeability of these vascular lesions. Further studies may prove useful in elucidating the life-history of specific lesions, as a guide to prognosis in individual patients and in assessing the response to various forms of treatment.